Background: At present, the prediction of adverse events (AE) had practical significance in clinic and the accuracy of AE prediction model after left atrial appendage closure (LAAC) needed to be improved. To identify a good prediction model based on machine learning for short-and long-term AE after LAAC.